- OPT RR is now a normal RRSet.
- Server stores only some EDNS options.
  (Will be changed later back to whole RR.)
- Rewritten EDNS getters and setters to use normal RRSet.
  (Option manipulation not done yet.)
- Added helper function for getting RRSet TTL.

This prevents problems when sharing/reusing the address bytes.

Zero-length msg_namelen prevents this.

 - Simpler DDNS signing, simpler changeset merging (todo: fix leaks)
 - Removed some needles helper functions and zone adjusting

Conflicts:
	src/knot/server/zones.c
	src/knot/server/zones.h
	src/knot/updates/xfr-in.c
	src/knot/zone/contents.c
	src/knot/zone/contents.h
	src/knot/zone/zone-dump.c
	src/knot/zone/zone-dump.h
	src/knot/zone/zonefile.c
	tests/process_query.c

Ttl Tests

ref #158

Ttl Check Fix

More info in the commit message.

Rollback Fixes

Some rollback calls were missing, I've added them, but I had to hack it a bit, there's a comment with explanation.

Previous logic was too complicated and confusing. Currently, there are
just two pregenerated source files which are selected on the basis of
the --disable-fastparser configure option.

Pregenerated files can be updated using scripts/update-parser.sh script.

- AXFR only
- no TSIG
- AXFR logging is not finished

no TSIG yet

- in DDNS, the check would not work if the node was previously somehow
  changed, the check in changeset application should take care of this.

- check in changeset application would not work because it was triggered
  only when adding RRs into empty node.

Use knot_rdata_t directly instead of via knot_rrset_t

Hostname acquisition